Deep Fake Detection Using Hybrid CNN and Machine Learning Techniques

Deepfake Detection is a critical task that involves the identification of manipulated media using advanced AI techniques. This is essential for preserving trust in the media's integrity and digital content's authenticity. Significant concerns exist regarding the potential misuse of faked images to generate misleading information, fake news, and malicious hoaxes. (CNN), (SVMs), and (KNNs) are used together in this study to create a more advanced framework for finding deepfakes. The proposed methodology begins with data collection from diverse sources, followed by preprocessing steps that include normalization and augmentation to enhance data variability. Researchers looked at the Hybrid of CNN, PCA, and the Shallow Face Net using accuracy and loss as metrics. The research showed that the combination of CNN, SVM, and KNN worked better with 93.5 % accuracy. AUC (Area Under the Curve) metrics significantly changed when we tried the model on different data sets. While the study shows that the framework is good at telling the difference between real and fake media, especially in low-quality compressed images, it also shows that it can be used for real-time deepfake spotting in many instances.